A Distributed Trust-based Reputation Model in P2P System

• Published in: Eighth ACIS International Conference on Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing (SNPD 2007) Vol. 1; pp. 294 - 299
• Main Authors: Yu-mei Liu, Shou-bao Yang, Lei-tao Guo, Wan-ming Chen, Liang-min Guo
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.07.2007
• Subjects: Analytical models; Artificial intelligence; Computer science; Digital signatures; Distributed computing; Feedback; p2p networks; Peer to peer computing; R-Chord; Reputation; Resists; Resource; Social network services; Software engineering; TBRM; Transaction; Trust
• DOI: 10.1109/SNPD.2007.478
• ISBN: 9780769529097; 0769529097

The P2P system is an anonymous and dynamic system, thus, some malicious behaviour can't be punished. In order to restrict the malicious behaviour in the P2P system, researchers have focused on establishing effective reputation systems. However, the present reputation system can't avoid the trick of the false reputation feedback. We propose a distributed trust-based reputation model in p2p system (TBRM) to avoid it. The TBRM algorithm differentiated the node's capability of providing honest quality by the nodes reputation value, and the honest evaluation by the trust value. In our model, the reputation value represented the resource quality of the provider, thus, other nodes would like this node have low reputation value. At this time, the false reputation feedback happened. In this paper, we used the trust value to restrict the false reputation feedback For the nodes with low trust value was difficult to get the required resource, we punished the false reputation feedback by low their trust value. We show by both theoretical analysis and simulations that the proposed TBRM algorithm can get quick convergence which is 11 times, high equity for both low and high reputation nodes, and can get a high successful rate of file- downloading. When the malicious nodes' rate is 80%, the proposed TBRM is about 95% successful rate, compared to the algorithm without reputation who is only 27%.